Entrepreneurial innovation requires the integration of skills and knowledge from multiple sources internal and external to the firm. Effective management of the network of external contributors requires the identification and selection of sources of knowledge and competencies required by the process, the integration of this knowledge along the process and the retention of the value of the generated knowledge. Preliminary findings from a longitudinal study of product innovation in six entrepreneurial ventures suggest that the likelihood of successful management of the knowledge flows in entrepreneurial ventures is increased by a solid base of related knowledge, by the existence of boundary-spanning roles, by a deep personal involvement in all the activities where knowledge is produced, by physical proximity with the partners, and by a careful design of the incentive system that links all the contributors to the learning process. The conceptual framework that is emerging from our study may represent a first step towards a learning-based theory of the entrepreneurial process that focuses on the underlying process of acquisition, production and retention of knowledge.

An established perspective in entrepreneurship studies contends that entrepreneurial activity relies on a network of supportive relationships for the acquisition of critical resources of different nature: money, goods and services, information, legitimacy, support, etc. (Aldrich & Zimmer, 1986; Birley, 1985; Larson, 1991; Lorenzoni & Ornati, 1988). Past literature on supportive networks has investigated how social networks influence the capacity of an entrepreneur to attract and make use of critical resources, focusing mainly on structural dimensions like size, intensity, diversity, and robustness (e.g. Dubini & Aldrich, 1991; Steier & Greenwood, 2000; Zhao & Aram, 1995). According to this literature, “the critical task confronting an entrepreneur (…) is to make use of his/her existing network of relationships and to elaborate and strengthen that network (Steier & Greenwood, 2000, p. 165).” In other words, entrepreneurs should build and shape their supportive networks, trying to optimize the structure of the network as a whole along the critical dimensions of range, redundancy, diversity, etc.

In this paper, we adopt an alternative perspective. We argue that the contribution of many external partners is essentially of knowledge-based nature. The development and realization of entrepreneurial ideas, in fact, requires a variety of data, skills, competencies, and technologies that are complementary to those possessed by the entrepreneurs. Industrial or research partners, suppliers, clients, consultants, venture capitalists offer a contribution that often goes beyond the “physical” content of the exchange: they contribute to the development and refinement of the technologies embodied in new products and services and of the organization that produces and delivers them. The success of an entrepreneurial venture, therefore, rests on an underlying learning process that draws on multiple external contributions. Effective management of this network of external contributors requires the identification and selection of appropriate sources of knowledge and competencies, the integration of this knowledge along the process, and the retention of the value of the generated knowledge. In this paper, building on findings from a multiple case study, we derive empirically-grounded theoretical propositions regarding the factors that affect the capacity of an entrepreneur to select sources of knowledge and skills, to make an effective use of them and to retain most of the knowledge generated in the innovation process.

Following a common methodology in inductive research, we will first present and discuss our guiding framework. We will observe how learning is an important, albeit underinvestigated, cognitive aspects of the entrepreneurial process. We will show how learning is complementary to opportunity recognition and risk-taking, in the general process that links the identification of an entrepreneurial opportunity, the decision to exploit it and the practical implementation of the innovative idea. In the following sections we will first present our research method and then we will discuss our findings in light of the conceptual framework proposed in section one. The general conclusions and implications for research and practice are discussed in the final section.

The fundamental learning process that underlies entrepreneurial activity, from the initial insight through various stages of development, is an issue still largely unexplored (Agndal, 1999). In the past, studies on entrepreneurial cognition have focused on two fundamental steps in the entrepreneurial processes: the discovery of opportunity (for a comprehensive review, see Shane, 2000) and the decision to exploit them, or, in other words, the attitude of individuals towards risk taking (e.g. Begley & Boyd, 1987; Brockhaus & Horowitz, 1986). Apart from some notable exceptions (Guth, Kumaraswamy & McErlean, 1991), little attention has been dedicated to what comes after—i.e. the learning process that takes place throughout the development and the realization of entrepreneurial ideas.

Learning in an entrepreneurial context seems to resemble what Miner and Mezias (1996) call generative learning, or discovery. Generative learning has a creative component that goes beyond repetition and incremental optimization: generative learning often involves the development of a completely new solution or a radically innovative product. As the task to solve does not involve repetition, the outcome of generative learning is not so much a change in routinary behavior, as a change in the interpretive frameworks, in the knowledge structures that sustain interpretation and action (Friedlander, 1983). In an entrepreneurial setting, this change is embodied primarily in a new product or service, in a new production process or in a new way to serve the market (Schumpeter, 1934). In this respect, we can conceive entrepreneurial learning as the process of building and refining a set of knowledge structures—technologies, routines, etc.—that expand and transform intuition into a viable new product or process. In other words, entrepreneurial learning tends to privilege the exploration of new possibilities, the experimentation of new ideas, the search for new alternatives, rather than the exploitation of existing competencies and resources and the refinement of current technologies (March, 1991).

A second distinctive feature of learning in an entrepreneurial context seems to be its collective nature. Although in an entrepreneurial context a single person—i.e. the entrepreneur—plays a pivotal role in the development process, this person rarely possesses all the competencies required for the success of the venture. Typically, the entrepreneur possesses a good knowledge of the market and the customers, and often a certain degree of technical competence in his field. This prior knowledge allows the entrepreneur to interpret new information as an entrepreneurial opportunity, leading to the generation of potentially valuable business ideas (Shane, 2000). The actual realization of these ideas, however, often requires skills and competencies that must be obtained from industrial, commercial and research partners, consultants, designers, etc. In an entrepreneurial context, then, learning often arises from the interaction of a number of actors that are in part external to the organization and extends from the entrepreneur—locus of coordination and impulse for the projects—and his close collaborators inside the company (technicians, engineers, marketing people, etc.) to a web of external partners, consultants and suppliers, who provide specific knowledge and competencies to the project.

Finally, from the entrepreneur’s point of view, learning takes place mainly at two levels. At a first level, learning is measured on the actual development of the desired technology, be it embodied in a product or a production process, at a reasonable cost and within a reasonable time frame. At a second level, learning is measured on the extent to which the entrepreneur himself has retained the novel knowledge that has been produced. Part of this knowledge is codified and part is tacit (Polany, 1966). The codified part is contained in the product design, in its technical specification and in its user manual, and its valuable insofar as it warrants a license. Sometimes, however, the tacit knowledge that is produced in the learning process is even more valuable. Often, besides the codified knowledge embodied in the product, an innovation project may lead to the development of specific—and largely tacit—competencies in new technological domains. These competencies may offer the basis for future development process that build on the knowledge accumulated in the past. In a world where competitive advantage is quickly eroded by competitors’ imitation and technological obsolescence, a solid competence base is a most valuable asset, in light of continuous renewal of the company’s product line.

In summary, therefore, an entrepreneurial context, learning seems to require the active management of a web of relationships in order to (i) identify and select sources of knowledge and competencies required by the process, (ii) integrate these ‘knowledge chunks’ along the learning process, and (iii) maximize the retention of the generated knowledge.

Our research was based on a longitudinal study of product or process innovation in six entrepreneurial firms, from the initial idea to the final version. General information on the companies and the projects that we observed is summarized in Table 1. All the analyzed cases are small and medium, privately-owned companies, and shared the same fundamental problem: the necessity to involve external contributions in the development of new products or production processes aimed at catching opportunities in the market. For each case, we followed the patterns of interaction between the company and the external partners as relationships were built and evolved, and contributions were acquired and integrated along the process. This method helped us capture the complexity of the learning process as it unfolded through the various phases of selection, acquisition and retention of knowledge, and to track the continuous evolution that social networks that sustain entrepreneurial action are subject to (Steier & Greenwood, 2000).

Data collection relied on different sources. A preliminary archival research (press interviews, web sites, etc.) helped us to collect background information on the companies. For each company, then, data were collected through semi-structured interviews with core internal and external contributors to each project. For each company we interviewed from 4 to 6 people, for a number of interviews ranging from 6 to 9 for each company. In all, 39 interviews were conducted with 28 persons (see Table 1). Each case study followed a standard protocol (Yin, 1989). The first interview with the main contact person, usually the entrepreneur or one of the partners, was aimed at (i) enriching our background information, (ii) identifying a suitable object of analysis for our research, and (iii) identifying the actors involved in the process and the nature of their involvement. In the concluding part of the interviews we asked our contact to indicate a number of persons that were highly involved in the process and that had a substantial interaction with external parties. These persons were then contacted for a second round of interviews. Interviews adopted an open-end format in order to collect both factual data and personal impressions. Our informants were first asked to reconstruct the story of the project as they lived it, trying to distinguish facts (how it started, who was involved etc.) from personal observations. We then asked to describe in more details, from their own personal experience, how the interaction with external contributors unfolded. To ensure reliability, both researchers were present at all the interviews. All the interviews were taped and transcribed. If information collected at a later stage required further probing or the clarification of minor discrepancies, some informants were interviewed more than once. Multiple interviews helped us move beyond individual perceptual biases, alleviate potential recall problems and reconstruct a reliable “story” of each process.

Data analysis used common methods for grounded theory building (Glaser & Strauss, 1967; Miles & Huberman, 1984) and combined within-case analysis to cross-case comparison. Within-case analysis, based on rich, at times anecdotal, information led to insights that were further developed and tested in cross case analysis. We began our analysis with a detailed reconstruction of the chronology of the process. We took note of the changes in the composition of the “activated network,” as external contributors joined the projects or terminated their collaboration. We then tried to highlight “critical events”—i.e. events or decisions that (i) involved a change in the composition of the activated network or in the characteristics of the relationships and (ii) marked an advance or a change of direction in the development projects. The identification of critical events was based on a content analysis of the interviews. We searched interviews for passages that contained references to the participation of the supportive network to the development process and implied a causal relationship with the advancement of the project. The analysis followed an iterative path, as often events or decisions that initially had only internal effects, later affected the dynamics of the supportive network. The search was conducted independently by the researchers; later comparison of independent analysis showed a substantial agreement. The literature on organizational learning and on networks offered us a terminology and a conceptual reference that helped us to relate each event (i.e. hiring Mr. Monteverdi, a young physicist from the University of Milano, Politecnico) to a more general theme (i.e. establishment of a boundary-spanning role). This coding procedure helped us to identify, for each case, a number of key themes. Each theme was then related to one or more of the three main learning issues (identification and selection of sources of knowledge, acquisition and integration of knowledge, retention of knowledge) according to the causal relationship expressed or implied by the informant. This analysis produced a number of causal relationships between “events” and “learning issues” that contained interesting insights on the management of knowledge flows throughout the network and along the process. Following indications from Eisenhardt (1989), we referred to the existing literature to develop and to enrich these inductively derived insights. Provisional interpretations and tentative propositions were refined in several iterations between theory and data until we were able to provide a plausible explanation of the observed patterns.

In a second stage, in order to verify how strongly each theme contributed to the explanation of the general phenomenon, we conducted a cross-case comparison. In some cases, the comparison required a further homogenization of concepts, as some themes (i.e. “perception of incentives,” “conflicting interests,” and “motivation”) were grouped into a more general concept (“alignment of interests and rewards”). At the end of this operation we were able to identify six themes that featured prominently across cases: “geographical proximity,” “alignment of interests and rewards,” “existence of boundary-spanning roles,” “possession of prior-related knowledge,” “personal involvement” and “socialization” (see Table 2). As it often happens in inductive research, these findings in part confirm and in part extend past literature. After having identified the key themes, cross-case comparison was used to verify the robustness of our provisional interpretations across cases. Again the process followed an iterative path, until the emerging conceptual framework fit the observed patterns across cases, and we were then able to relate each theme to one or more learning issues (see Table 3).

As anticipated in the method section, the explanation of how the observed learning processes unfolded relies on different combinations of six fundamental factors. In this section, using some evidence from the cases, we will illustrate and discuss how these factors can improve or reduce the effectiveness of the processes of selection, acquisition and integration, and retention of knowledge.

Identification and Selection of Sources of Knowledge

Our analysis showed that the effectiveness of the identification and selection of external sources was affected by factors related in part to the process and in part to the criteria used. As far as the process was concerned, the identification of the potential contributors followed two main routes. At first, when confronted to a task to be carried out by an external partner, most entrepreneurs tended to rely on existing relationships and started searching through their personal network. Mr. Tosi, for instance, relied on a machine shop that had worked for him before. At Futureplast, Mr. Guzzoletti contacted an engineer from a research company that had visited the company months before. Parma consulted the managing director of a commercial partner. Polti relied on established relationships with external laboratories and suppliers of components. In this respect, we could talk of a process of “network activation,” as latent relationships were accessed and external contributors were harnessed along the specific project. Past literature on entrepreneurial networking observed how the size of a personal network helps reduce the costs associated with the search and the selection of transaction partners (Dubini and Aldrich, 1991). This process of “network activation” also helps overcome the problem of assessing the potential quality of the contribution, because often the required competencies belong to a scientific or professional domain of which the entrepreneur has little knowledge or experience. As a manager from Parma put it: “It’s difficult to assess and control competencies that you do not have (…) that’s why it is important to know your partners in advance.”

If the personal network did not offer any suitable solution, they started a rational search, based on the analytical comparison of potential suppliers of a specific product or service. Data about existing alternatives were gathered extensively. A restricted range of potential partners was selected and contacted. The nature of the project and the content of the contribution were discussed, further information was gathered, until one of the potential partners was selected. The observed entrepreneurs, however, did not always possess the required information and skills neither to identify suitable candidates, nor to perform an appropriate selection. In these cases, we observed a tendency to rely on what we could term “boundary-spanning roles” that helped the entrepreneurs reduce the uncertainty associated to the decision. For instance, in entrepreneurial ventures that built on scientific knowledge platforms and required the performance of activities of scientific nature (lab tests, trials, etc.), internal scientists provided the requisite knowledge to identify, contact and select academic institutions. This, for instance, was the reason why the increased complexity of the development activities carried out at Futureplast required the employment of a skilled physicist, whose employment, as Mr. Guzzoletti, owner of Futureplast, put it, “was meant to provide structure and method to the research activity carried out at Futureplast, and to open up new channels towards the external acquisition of new, front-line scientific knowledge.”

Boundary spanning roles, however, could be found also among the personal network. In the cases of Microalgae, Parma and Petroltecnica, the projects required the contribution of technical and professional skills of which the entrepreneurs had no previous experience. In all the cases, the selection and identification of appropriate contributors was essentially delegated to third parties, members of the personal network of the entrepreneur that were involved in the project not only for their personal skills, but also for their knowledge and the connections with other networks belonging to different technical and professional domains. As Parma decided to diversify from simple safety boxes to more complex cash dispensers, for instance, the first move was to involve in the project Mr. Spinetti, a former commercial partner with years of experience in running companies in the cash dispenser industry. Mr. Spinetti did not contribute only with his knowledge of the market: he personally conducted the selection of the technicians to be contracted or hired to supplement Parma’s lack of specific knowledge about cash-dispenser design. The selection was conducted among former employees and partners of Mr. Spinetti.

Cohen and Levinthal (1990) observed how the ability to evaluate and utilize outside knowledge depends on the level of prior related knowledge possessed by the company. Related knowledge includes basic skills, a shared language and up-to-date knowledge of the most recent scientific or technological developments in a given field. In this respect, boundary-spanning roles provide the related knowledge that entrepreneurs often lack, especially when they approach new scientific or technological domains. Furthermore, boundary-spanning roles help extend the search from the personal network of the entrepreneur to a broader network that includes even their own—what Dubini and Aldrich (1991) call the “extended network”—thus reducing the time and resources invested in the collection of information about a specific domain. In summary:

Proposition 1a: The existence of boundary-spanning roles positively affects the entrepreneur’s capacity to select and identify external sources of knowledge.

As far as selection criteria were concerned, most entrepreneurs looked first at the competence level of the prospective partner. This criterion, however, did not always lead to success. When looking for a partner to develop the support truck for its tank-cleansing robot, Petroltecnica’s first choice, for instance, fell on one of the leading producers of tank-cleansing trucks, a company located in Pordenone—more than 300 km from Petroltecnica’s headquarters in Rimini. However, after a month of loose interaction during which Petroltecnica obtained few or no replies to the numerous designs and plans submitted to the partner, the entrepreneur reneged on the contract and looked for a closer working partner. According to him, the distance impeded the establishment of a cooperative spirit required for a timely and effective completion of the project, because “the physical distance between their team and ours hampered the creation of the basis required for them to understand our needs here at Petroltecnica.” Mr. Pivi later chose two companies located much closer to the headquarters, the collaboration with whom proved to be much more effective. The unsuccessful experience of Petroltecnica, however, seems to indicate the importance of another factor: the specific interests of the external partners. In fact, our informants shared the impression that the first company that they contacted, a large-scale producer, was only marginally interested in the project as it required the customized development of a product that was destined to be produced in small numbers. Therefore, the project was assigned a low priority. On the contrary, the second company that they contacted was a design and engineering firm, whose core activity lied in the development of custom-made new products and technologies. After a thorough discussion on the objectives and the requirements of the project, the process quickly got started. The influence of this variable was even more apparent in the case of Futureplast, where the comparison of a failed and a successful project showed how one of the discriminating variable between the two process was the absence of an incentive for a critical contributor in the preliminary stage of the failed project. In this case, the reproduction of an existing technology required the collaboration of CNR, an academic research laboratory with no commercial drive. Even from an academic point of view, CNR had no specific interest in the specific object of research—fiber optics—and two of our informants expressed the opinion that CNR did not even seem much interested in investing on the development of specific competence in the field. CNR considered the project as just another external research order that they fulfilled diligently, but with not particular commitment. The project never went beyond a preliminary stage. Another project conducted by the same company—the development of a laser system—had a completely different story. The critical contributor in this case was CISE-ENEL, part of a recently-privatized state-owned electric supplier, whose mission had recently been re-framed, as scientists and technicians were encouraged to look for external partners or clients, in order to find a profitable application for the capabilities of the center. CISE-ENEL came in contact with Futureplast, when Mr. Nava, an engineer from CISE, was looking for new clients for the capabilities developed in the laser technology over thirty years of activity. Furthermore, Nava had a personal interest in the project as he had already been attracted by the technology even before getting in touch with Futureplast. In summary, our analysis suggests that the identification and selection of external sources of competencies benefits from the existence of boundary-spanning roles and should consider not only the competence, but also the motivation and the geographical location of the prospective partners, as the latter seem to affect the process of integration, as we will discuss in the next section. In more formal terms:

Proposition 1b: Other things being equal, the inclusion of geographical location and the degree of interest in the project within the selection criteria increases the likelihood of success of the relationship.

Acquisition and Integration of External Knowledge

Some of the factors that improved the effectiveness of the selection process had also an impact on the capacity of the entrepreneur to acquire and to integrate external knowledge along the innovation process. Boundary-spanning roles, for instance, did not only help select external sources of knowledge, but also facilitated the subsequent interaction. In Futureplast, for instance, in the initial stage of the laser project, Mr. Monteverdi, a newly-hired physicist, was assigned the task to collect and review the related scientific literature, in order to learn the fundamentals of laser technology. A comprehensive library was set up in Futureplast laboratories, after having thoroughly searched local research centers. During this stage, the presence of Monteverdi helped Mr. Guzzoletti to acquire the basic terminology and the fundamental notions of physics that let him play an active role in the design of the system and of the various components. When Microalgae enter a later stage of development and, from a small research lab where external academics conducted experiments in their spare time, became a research company with a range of external contributors, permanent staff was selected on the basis of their capacity to interact with these contributors, rather than on their previous experience in the business. The employment of Mr. Carella, an agronomist, and Mr. Lupoli, an electronic engineer, marked a substantial advance in the project, as Microalgae was finally staffed with persons that could act as a nexus for the contributions of scientists and researchers from different universities, and for the suppliers of the various technologies required by the project. Until then, the only permanent employee of the company was little more than a guardian for the prototype plant, while the contribution of Mr. Gregorini, the entrepreneur, did not go much farther than the initial idea and a periodic supervision. Also, boundary spanners may contribute to legitimate the entrepreneur in the eye of academics, helping overcome occasional reluctance to be involved in a commercial project. As Mr. Guzzoletti from Futureplast told us: “At first, universities were reluctant to cooperate with me, because I did not have an academic degree, but I had built my knowledge base on practical experience. The presence of Monteverdi was important (…) in overcoming mistrust and in the establishment of a relationship.” In summary, evidence from the cases suggests that boundary-spanning roles that provide access to other professional, technical or scientific domains, contribute to the acquisition of the knowledge base required to lead the process of knowledge integration. In particular, internal scientists may provide not only the language, but also the legitimation required to interact with academic institutions and profit from their collaboration. In formal terms:

Proposition 2a: the presence of boundary-spanning roles positively affects the acquisition and integration of distributed knowledge along the innovation process.

In the last section, we observed how the geographical location and the interests of the prospective partners should be included, along with the competence level, in the selection criteria. In fact, as we have anticipated, the reason why these factors should be considered is that they seem to affect the efficiency and the effectiveness of the acquisition and integration of knowledge. As the required knowledge is distributed across a network of independent actors, successful learning requires that they all share an interest in the successful completion of the project. In the last section, we have already observed how in part this motivation may be provided by the goals of the contributors and by the nature of their involvement. The design of the contractual system that frames the relationship with the external partners, however, should contribute to realign diverging interests, providing or reinforcing incentives for all the parties to contribute with their skills and competencies to the timely and efficient completion of the process. In the case discuss before, the contract between Futureplast and CNR did not help compensate the latter’s lack of interest in the project, as the compensation was not tied to the achievement of any objective, but only to the carrying out of a series of tests. In the laser project, besides CISE’s and Mr. Nava’s intrinsic interest in the project, for instance, CISE’s efforts were also stimulated by the form of the contract, the periodic renewal of which was tied to some efficiency objectives, measured over the advancement of the process and the respect of programmed milestones. Similarly, the relationship between Petroltecnica and its major partner, RB1, was regulated in details by a contract that specified for each stage, the responsibilities of the contractor and the customer as regards, for example, the sharing of information and the respect of deadlines. Each contractor was given the possibility to renege in every stage if the counterpart would not collaborate according to the norms. Parma went even further, asking Mr. Spinetti, a major partner in the project, to set up a company, Eurolab, to share the risk of the enterprise. Most technicians were offered a term contract that would be made permanent only after the preliminary stages had demonstrated the industrial and commercial viability of the new product. In fact, Parma did not internalize the core of the project, until a relatively late stage of development. In summary, evidence from the cases suggests that (i) a limited, even indirect, participation to the entrepreneurial risk and to the potential rewards of the project, and (ii) a clear definition of responsibilities for each stage of the process, positively affect the efficiency and the effectiveness of the process.

Proposition 2b: the explicit connection of rewards and punishments with the successful and timely completion of the project positively affects the acquisition and integration of distributed knowledge along the innovation process.

Similarly, change of geographical location was often associated with substantial advancements in the projects. In the cases of Microalgae and Petroltecnica, for instance, decisive steps in the development process were made when Mr. Gregorini and Mr. Pivi respectively decided to move research and development activities closer to the headquarters. In the first case, research did not make much progress until Gregorini moved the labs from southern Italy to Switzerland, where he could personally supervise the development of the project. Until then, Microalgae had been little more than a place were three researchers from the National Council would periodically pass to test the results of experiments that were mainly conducted elsewhere. In the second case, as we have already mentioned, Mr. Pivi, who had initially contracted two companies on the basis of their distinctive competencies, was later forced to change partners, and chose two companies located in Ferrara and Rimini. The collaboration with them proved to be much more effective, as the reduced distance allowed frequent meetings and facilitated the establishment of a mutual understanding that our informants considered fundamental for the successful cooperation. Evidence from the cases, then, suggests that reducing geographical distance between the parties (i) increases the possibility to control and provide impetus to the process and (ii) facilitates the development of cooperative relationships between the parties. In formal terms:

Proposition 2c: Geographical proximity positively affects the acquisition and integration of distributed knowledge along the innovation process.

As we have mentioned in the last section, past studies (Cohen and Levinthal, 1991, Szulanski, 1996) shown that the amount of knowledge possessed by an organization or an organizational unit influences its capacity to acquire related knowledge from another organization or unit. Prior related knowledge is expected to increase the capacity to appreciate the value of external knowledge and to facilitate its understanding and subsequent adoption. Although we had the impression that all the observed companies possessed a core set of skills and expertise, that constituted their distinctive competencies, only in four of them, Polti, Parma, Microalgae and Futureplast, the presence or the gradual development of such a knowledge base was explicitly connected by our informants to the advancement of the projects. For Polti and Parma, this knowledge base resided in part in a profound knowledge of the market, and in part in specific technical competencies in the design of home appliances of small size, in one case, and of safety devices, in the other. In both cases, these competencies formed the backbone of the projects, because they concerned the architecture of the products. External contributions were called in to develop specific components— such as the engine and the philter of the vacuum cleaner, or the CPU and the banknote sorter of the cash dispenser—according to the indications of the two companies, and later assembled on the architecture designed internally. Where the impact of the knowledge base was more evident, however, was in the case of Futureplast, where the presence of a strong knowledge base emerged as one of the factors discriminating between a failed and a successful project. In the first case, although the potential application of the technology had a closer link to the core business of the company, the development of the new material required fundamentals of optic and chemistry that were far from anything that Mr. Guzzoletti and his staff had mastered before. On the contrary, in the second case—the development of a laser system—at the start of the project Mr. Guzzoletti and his assistant, Mr. Monteverdi had already amassed a considerable body of knowledge in a specialized library. In this phase, the role of Mr. Monteverdi had been fundamental, not only as a “broker” of books and scientific contacts, but also because he had introduced Mr. Guzzoletti to the basic terminology and concepts of physical laws, providing him with a sufficient knowledge base to understand the nature and implications of the basic technologies of the systems. In this case, Mr. Guzzoletti could retain a direct control of the process, while in the first case he eventually had to delegate most responsibility to an external partner, CNR, whose commitment to the project, as we have already mentioned, was marginal. In formal terms:

Proposition 2d: Prior related knowledge positively affects the capacity of the entrepreneur to manage the integration of distributed knowledge along the innovation process.

As emerged in the cases of Polti, Parma and Microalgae, the successful integration of all the external contributions required a redesign of the components and the architecture of the product or the process in order to improve the performance of the system and the fit between all the parts. In turn, this required the partial acquisition of skills and competencies of external partners. This knowledge, however, was largely tacit. Part of the technical knowledge involved in the design of vacuum cleaners or banknote sorters, for instance, it is not codified in manuals, models or procedures, but it is the product of experience. As our informants explicitly declared, if a company with no previous experience in the field wants to learn how to do it in a reasonable time and at an affordable cost, the only way is to hire somebody with experience that can transfer that knowledge to the company. The transfer of knowledge, however, was not conducted through formal training, but as external partners and internal staff—sometimes the entrepreneur himself—worked together. In the case of Polti, for instance, as vacuum cleaners were new to the company, they temporarily hired a retired engineer that could assist the company in the initial development of the first line of products. Parma asked external technicians involved in the development of the first prototypes of the banknote sorters to move in the company’s labs. Evidence from the cases, then, shows that the acquisition of external tacit knowledge can be facilitated by creating the conditions for external partners and internal personnel to work together on experiments, design, etc. This finding is congruent with past research that has shown how the transfer of tacit knowledge requires a process of socialization (e.g. Nonaka, 1994). In formal terms:

Proposition 2e: Intense socialization positively affects the capacity to acquire and make use of external tacit knowledge along the innovation process.

Retention of New Knowledge

A critical issue for entrepreneurs, is not only the successful development of a new product or process, but also the appropriation of the full value of the set of knowledge structures that constitutes the result of the innovation process. As we have observed earlier, part of this knowledge is tacit and part is codified. In the Microalgae case, for instance, besides the basic design of the photobioreactor that was eventually licensed, what really made the company competitive was the experience gained in the management of the production facility: although a competitor could easily replicate the basic ideas behind the new production process, in fact, its effective management requires a intimate knowledge of how the various variables affect the ecology of the micro seaweed— an intimate knowledge that is the result of the development process itself. In other cases, such as Parma or Futureplast, besides the codified knowledge embodied in the product, the development process led the companies to acquire specific—and largely tacit—competencies in new technological domains. All the entrepreneurs that we observed, therefore, patented their new creations, but often they also asked internal and external contributors to keep track of the progress made, by formalizing routines, procedures, and manuals that converted experience into hard information. This helped them to retain an understanding of the steps and the ways in which advancements were achieved. Yet this did not let them capture hundred per cent of the outcome of learning. The soft side of knowledge, what people possess because they do, experiment and repeat the action, cannot be completely codified. Evidence from the cases suggests that a critical factor for the appropriation of tacit knowledge was a deep and direct personal involvement of the entrepreneur in the project. At Futureplast, most of the research activities that led to the successful development of the laser system took place at the company’s labs, while Mr. Guzzoletti was working in close contact with the external technicians. In this way, Guzzoletti was able to observe and discuss their tests and trials. Reverse engineering was systematically done on all the components of the system in order to acquire all the relevant knowledge, especially on the electronic components and on the laser source, so that the company could, later, reconstruct autonomously the whole system. While intense socialization helped some companies to acquire part of the existing skills of the external partners, direct involvement in development activities facilitated the appropriation of the newly created knowledge. In formal terms:

Proposition 3a: Personal involvement in the process positively affects the entrepreneur’s capacity to retain the tacit knowledge created in the innovation process.

Across the six cases, however, we observed significant differences in the degree of involvement of the entrepreneurs. Mr. Guzzoletti and Mr. Pivi, for instance, personally followed the whole process, while in the case of Microalgae, Mr. Gregorini left considerable autonomy to the experts he gathers, since he did not have at all the necessary competencies to understand what they needed in order to develop the product. Again, what seems to have made the difference was the possession of prior related knowledge. Entrepreneurs with a technical background displayed a higher involvement in the process as they took an active role, continuously offering suggestions, bringing new ideas and at times intervening even in the solution of minor technical problems. In other cases, all the designing and experimenting was left to the technicians, while the entrepreneurs periodically intervened to keep the development process oriented towards the market. These comments and suggestions tended to concentrate on the commercial side of the product, its congruence with the characteristics of the customers and the distribution. In these cases, however, the benefits reaped by the entrepreneur did not go much farther than the commercial value of the product or the license. In summary:

Proposition 3b: Prior-related knowledge positively affects the capacity of the entrepreneur to retain the new knowledge produced in the innovation process.

The ability to continuously learn and generate new knowledge is one of the key conditions for the sustained success of entrepreneurial ventures. From a cognitive point of view, however, entrepreneurial activity is usually portrayed either as risk taking or as opportunity recognition. These perspectives tend to overlook the individual cognitive processes and the collective interaction that take place as opportunities are explored, solutions are developed, and their potential risk and returns are gradually made sense of. In this paper we reported findings from a study of factors that affect the effectiveness of learning processes in entrepreneurial ventures. Our findings suggest that possessing a solid knowledge base, being deeply involved in all the activities (from testing and prototyping to after-sales service) where learning potentially takes place, maintaining a physical proximity with the partners, and carefully designing an incentive system that links all the contributors to the learning process are all factors that increase the likelihood of a successful production and retention of new knowledge. Our evidence, then, seems to suggest a sort of learning-based theory of entrepreneurial activity, according to which an entrepreneur at the center of a “learning network,” should directly perform, or at least keep a high degree of involvement, in all the activities where learning takes place, as contracting out these activities means reducing the amount of valuable knowledge appropriated in the process, and losing control of them may reduce the pace and effectiveness of learning. In particular, we observed that the criticality of the retention of new knowledge grows as larger and larger valuable fractions of this knowledge fall outside the codified domain. The fact that other actors, internal or external, retain this tacit knowledge poses a serious question of the possibility for the entrepreneur to appropriate the whole value arising from the learning process. Although the property of the license helps the entrepreneur to seize the value of the codified portion of the newly created knowledge, how to retain the tacit portion is an open problem.

The study was part of a larger, multi-partner research project financed by the European Union and termed “Small Business Training and Competitiveness: Building Case Studies in Different European Cultural Contexts.” Although the paper reports findings and interpretations that reflect solely the specific approach of the local research team, the comparison with our partners’ findings helped us test and refine our emerging framework.

CONTACT: Davide Ravasi, Strategic and Entrepreneurial Management Department, SDA Bocconi—Bocconi University School of Management, Via Bocconi 8, Milano, 20136 Italy; (T) +39-02-5836.2540; (F) +39-02-5836.2530; davide.ravasi@uni-bocconi.it

Agndal, H. (1999) “Individual Learning Among Entrepreneurs—Towards a Research Agenda.” in Entrepreneurial Knowledge and Learning. Conceptual Advances and Direction for Future Research, C. Salvato, P. Davidsson & A. Persson (Eds), 48–63, JIBS Research Report, Jönköping.

Aldrich, H. & C. Zimmer (1986) Entrepreneurship through Social Networks.” In The Art and Science of Entrepreneurship. D. L. Sexton and R. W. Smilor (eds.), 3–23. Cambridge, MA: Ballinger.

Begley, T. & D. Boyd (1987) “Psychological Characteristics Associated with Performance in Entrepreneurial Firms and Small Businesses.” Journal of Business Venturing 2: 79–93.

Birley, S. (1985) “The Role of Networks in the Entrepreneurial Process.” Journal of Business Venturing 1: 107–117.

Brockhaus, R. & P. Horowitz (1986) “The Psychology of the Entrepreneur.” In The Art and Science of Entrepreneurship. Donald Sexton & Raymond Smilor, eds. Cambridge, MA: Ballinger.

Cohen, W. & D. Levinthal (1990) “Absorptive Capacity: A New Perspective on Learning and Innovation” Administrative Science Quarterly 35: 128–152.

Dubini, P. & H. Aldrich (1991) Personal and Extended networks are central to the Entrepreneurial process.” Journal of Business Venturing 6: 305–313.

Friedlander, F. (1983) “Patterns of Individual and Organizational Learning.” in Paul Srivastava and Associates (Eds.) The Executive Mind: New Insights on Managerial Thought and Action. San Francisco, CA: Jossey Bass.

Glaser, B. & A. Strauss (1967) The Discovery of Grounded Theory: Strategies for Qualitative Research. Chicago, IL: Aldine.

Guth, W., Kumaraswamy, A. & McErlen, M. (1991) “Cognition, Enactment and Learning in the Entrepreneurial Process.” Frontiers of Entrepreneurship Research, Wellesley, MA: Babson College.

Larson, A. (1991) “Partner Networks: Leveraging External Ties to Improve Entrepreneurial Performance.” Journal of Business Venturing 6(3): 173–188.

Lorenzoni, G. & O. A. Ornati (1988) “Constellations of Firms and New Ventures.” Journal of Business Venturing 3: 41–57.

March, J. G. (1991) “Exploration and Exploitation in Organizational Learning.” Organization Science 2 (1): 71–87.

Miles, M. & A. Huberman (1984) Qualitative Data Analysis. Beverly Hills, CA: Sage.

Miner, A. S. & S. J. Mezias (1996) “Ugly Duckling No More: Pasts and Futures of Organizational Learning Research.” Organization Science 7: 88–99.

Nonaka, I. (1994) “A Dynamic Theory of Organizational Knowledge Creation.” Organization Science 5: 14–37

Polany M. (1966) “The Tacit Dimension,” in Prusak (1997) Knowledge in Organizations, Butterworth-Heinemann.

Schumpeter, J. A. (1936) The Theory of Economic Development, Second Edition. Cambridge, MA: Harvard University Press.

Shane, S. (2000) “Prior Knowledge and the Discovery of Entrepreneurial Opportunities.” Organization Science 11: 448–469.

Steier L. & R. Greenwood (2000) Entrepreneurship and the Evolution of Angel Financial Networks.” Organization Studies 21 (1) 163–192.

Yin, R. K. (1989) Case Study Research: Design and Methods. Newbury Park, CA: Sage.

Zhao, L. & J. D. Aram (1995) “Networking and Growth of Young Technology-Intensive Ventures in China.” Journal of Business Venturing 10: 349–370.